The application of pattern recognition technology in traditional Chinese medicine

摘要

The Chinese materia medica was classified into several categories by experience. It is difficult to find the material basis of this classification only by chemical ways. As near infrared spectrum can provide internal molecular information of material, it can be used to analyze the ingredients of medicines. Firstly, the NIR spectra of six kinds of medicines in the same category were used to make sure that this method can identify the medicines correctly, and then we use the spectra of medicines belonging to different categories and combined with LSA technology to classify the medicines. The cosine similarity was used to calculate the correlation. The result was transformed to a visualization of the correlation matrix and was then shown by a gray-scale picture. The classified result turns out to be perfect.